‘How long did it take to get here?’
Data Collected to Test Hypothesis 3: For this hypothesis, I collected two pieces of data. Firstly, I obtained three bus maps for Kingstown, Surbiton and New Malden. By using these maps (secondary data) I counted the number of bus routes which passed through all the areas. Secondly I used a questionnaire, which had the following question also necessary to test this hypothesis:
‘How did you get here?’ This will let me get a brief understanding of how many shoppers used the bus or rail services to get to the shopping area they visited.
Method Used to Collect the Data: To collect the data that was required for the coursework investigation, all the three Year 11 geography sets were given an area to which they would go to- Kingston, Surbiton or New Malden. From these areas, classes were divided into 9 shopping sites. Our class was split into three groups to collect data in Surbiton, Surbiton Park Parade and Alexandra Drive, where my partner and I were stationed. We were each provided with a questionnaire (primary data) which contained 5 questions, and as shoppers provided us with the answers, we would fill out the questionnaire.
But before we began to ask people for the information required for our questionnaires, we performed a shop count and recorded the type of shops in our given area, which conveyed much valuable information about the area and also its shopping traits. This in turn would help categorise the shopping centre in some sort of hierarchy. Some areas contained a range of large shops selling high order goods whilst others contained less important small convenience shops. And due to this, not all groups had been evenly split up as not many customers would be shopping at small low order convenience shops compared to larger high order shopping centres (diagrams can be seen on the following page).
The questionnaire, which supplied us with results of extremely valuable information required for our coursework, needed to be reasonably short. This ensured that more data could be collected by a range of people rather than more time being wasted on just a few. With a short questionnaire, mostly with more closed answer questions rather than open answer questions, we were guaranteed with more co-operation from the shoppers. The questions were laid out in a polite manner and responses by the shoppers were recorded in the appropriate columns on the questionnaire.
Once all the data from every shopping site in the three areas had been collected, the data was pooled altogether in a single table on a database. The data which had been gathered at each shopping area was a representative sample of the population in the different areas. Therefore this meant that the data collected may not be true for the whole population in the different shopping centres.
Limitations: When carrying out the investigation, we did acquire some accurate results; however these may have been altered if the data collection took place on a Saturday and not on a weekday. We encountered many working people on the Thursday and as a result they did not have any time to answer our questions. Also as our data collection was close to school ending hours, many parents were rushing to pick their children up from school and some were even just nipping out instead of shopping. Another limitation to the investigation was the weather. It was bright and sunny on the Thursday we went, but if it was raining or really cold weather, then this would have altered our results extremely. Being on Alexandra Drive housing just a few shops, we were lucky to get a reasonable amount of people to answer our questionnaires, but quite a few of the shoppers were not really that interested and could not be bothered and this made the data collection even more harder. One other limitation was operator variants. Some people did not collect data in the same way that we were supposed to and therefore their results were not useful for the investigation (e.g. some people did not get road names but just a specific area). Also, without realising, the sample may have become very selective as to where the data came from and I and my partner experienced this, as we saw members of the public who we thought would not want to answer our questionnaires. This would provide us with bias data as we may have been likely to ask individuals who were of a similar age, gender or dress sense to us.
Anomalies: From my results, I can see that most of the data is comparable, except I did notice one anomaly from it. The average time it took shoppers to get to where they were shopping was 7 minutes for the reason that one shopper had come all the way from Chessington, which took the shopper half an hour. This is an anomalous result as the average time would be just 4 minutes providing that we did not ask this shopper questions for our questionnaire.
By looking at the ‘classification of shops and services’ sheet, I also spotted one anomaly from this. When performing the shop count in Alexandra Drive, we encountered two shops and services classified under ‘car sales and services’, which retail in very high order goods. This would make the data quite difficult for the reason that car sales and services tend to be outside of the main shopping centre, as a large amount of space is required and they would therefore not be willing to pay a lot of money to display their cars in a large shopping centre. Instead they would set up in a smaller settlement where there is less cost for land. Cars are high order and thus these can be found in smaller settlements which would distort Christaller’s theory.
Raw Data:
Data Presentation- section 3
Before I begin to use various graphical techniques to determine whether my hypotheses are proved or disproved, I decided to rank the 9 different shopping centres according to how many shops there are in the shopping centres, as this is the main basis to all my hypotheses and it is now easier to compare the shopping centres too. After I ranked them, I positioned them into three Tiers (T1, T2 and T3). Tier 1 will be for shopping centres with 200 or more shops, Tier 2 for those with 100 or more shops and Tier 3 for shopping centres with less than 100 shops.
Having ranked all the shopping centres in table 3, I produced a column graph (on page 17) showing the amount of shops in each area, which immediately shows a hierarchy between shopping centres in the borough of Kingston. The column graph shows a clear comparison of all the shopping centres in the borough.
On page 18, a stacked column graph can be seen showing the types of shops in each of the nine the shopping centres in percentages. It shows that the larger shopping centres have a wider range of shops and that they have more shops selling high order goods. From the stacked column graph, by showing the data in percentages and using columns of the same size, the different types of shops are easily comparable.
Graph 1
Graph 1 shows the number of shops in each of the nine shopping centres and from looking at the graph, the data is easily comparable. By far, for tier 1, Kingston Town Centre is the largest shopping centre containing 572 shops. The second biggest shopping centre, Surbiton, had 403 fewer shops consisting of 169 shops and the third largest is New Malden High Street with 143 shops, both of which come under tier 2. The next tier of shopping centres have a lot fewer shops than those in tier 1 and 2, with Burlington Road and Surbiton Park Parade both consisting of 30 shops and Kingston Road 1 with one shop more. Finally, Alexandra Drive has only 18 shops with The Triangle being the smallest shopping centre with only 11 shops.
Graph 2
Graph 2, showing the types of shops in each shopping area, depicts an expected trend that the larger shopping centres contain a greater number of different types of shops. Kingston Town Centre, Surbiton and New Malden High Street all include the nine different types of shops whereas the smaller shopping centres such as Alexandra Drive have only 4 types of shops and The Triangle, surprisingly with 5 types. Graph 2 shows that a wider range of shops are found within the larger shopping centres. This is for the reason that smaller shopping centres, such as those on Alexandra Drive, support a smaller threshold population, whereas larger shopping centres, such as Kingston, support a much larger threshold population because they stock a wide variety of high order goods, which inevitably attracts a large number of people for all different purposes. Hence the business in the area grows overtime matched by an increase in threshold population.
By looking at graph 2, I can see that there are a higher proportion of convenience shops in smaller shopping centres than others. Only 2.5% of the shops make up convenience shops in Kingston, whereas the smallest shopping centre, The Triangle, consists of almost 40% of these. This indicates that as the size of the shopping centre decreases, the proportion of convenience stores increases, although it is evident in graph 2 that the proportion of convenience shops fluctuates.
Department and variety stores selling high order goods are only found in Kingston Town Centre, Surbiton and New Malden shopping centres because these types of shops require a larger threshold population to maintain the shop and make a profit. By being located in the larger shopping centres, ensures that the threshold population is greater as the sphere of influence is greater for the larger shopping centres (table 4). It can also be seen that the number of car sales and services are in greater proportion in the smaller shopping centres, which will increase the size of their spheres of influence as shoppers will be willing to travel a further distance for a high order good.
Hypothesis 1
A shopping centre which contains the greatest number of shops will have a large sphere of influence
For this hypothesis I have created a scatter graph (see graph 3) using the figures from table 4, below which shows the number of squares for the size of sphere of influence for each shopping centre. On the following page, table 5 and a column graph (see graph 4), can be seen showing the size of sphere of influence for the names of the different shopping centres.
Hypothesis 2
The amount of time it takes to travel to a shopping centre with many shops will be greater than travelling to one with not so many shops.
To show how long it took shoppers to reach the shopping centres with different numbers of shops, I have used a scatter graph below which has time measured in seconds.
Hypothesis 3
More shoppers will use public transport methods if travelling to a shopping centre with many shops.
To present the data for this hypothesis, firstly a simple line graph below can be seen which shows the highest number of bus routes first and then it slowly descends to the lowest number of bus routes in the shopping centres.
However on the following page, is a scatter graph (see graph 7) showing the number of shops against the number of bus routes in a shopping centre. There is also a column graph (see graph 8) to show the number of train stations in some of the shopping centres.
BUS MAP FOR KINGSTON
BUS MAP FOR SURBITON
BUS MAP FOR NEW MALDEN
Below is a stacked column graph and from it, you can see the number of shoppers whom either came by bus, rail, or none of these public transport methods were used. Graph 10 is simply a scatter graph to show the number of people who used public transport to get to shopping centres with various numbers of shops in them.
For the data collection section, me and my partner managed to take digital pictures of the different shops in the shopping centre.
Graphical Techniques Grid
Data Analysis- section 4
In the fourth section of the investigation, I will be analysing and interpreting the data, included in the data collection section, to identify any patterns and relationships which are evident. I will be including simple statistical tests, such as the Spearman Rank Correlation test and calculating the Semi Average Regression line. Firstly I will be writing about how the spheres of influences for the shopping centres were found, how we drew on the semi average regression lines and lastly how the Spearman Rank Correlation was also calculated from the data.
Method Used to Find the Sphere of Influence
Having found out where shoppers came from by using the questionnaires, we were able to construct a desire line map with lines showing the areas shoppers came from to the centre of the shopping sites. However due to operator variants, as some people may not have been there to shop, some lines would have been extended out too far and would then show anomalous results.
Subsequently, to plot on the sphere of influence for each shopping centre, tracing paper was placed over all the various lines, represented by different colours for each shopping centre, to trace over the desire line map and then to draw on the spheres of influences which could have been manipulated for the spheres to appear best. Once the 7 different spheres of influences had been drawn on the tracing paper, it was then placed on top of squared paper in order for the number of squares to be counted within the spheres.
Method Used to Calculate the Semi Average Regression Line
Firstly a scatter graph must be plotted and a positive trend should be identified. Then the average of the data for variable x and variable y (always number of shops) must be found. Having calculated the averages for both variables, they should then be plotted further on the scatter graph. However, since the semi average regression line needs to be plotted, all the values on the x-axis and y-axis which are greater than the x-average and y-average need to be found. This is done by adding up all the values from both axis and then dividing them by how many values are bigger than the x-average and y-average. The final set of values which have been calculated are then plotted on the scatter graph too. Finally, the above method is used once more, except this time the values on the x-axis and y-axis which are smaller than the x-average and y-average must be found.
Method Used to Calculate the Spearman Rank Correlation Coefficient
Ensure that there are two relevant variables which can be tested for significance. Once the two variables have been chosen, a table similar to the one evident on page 48 onwards should be created. Then rank the two data sets and this is achieved by giving the ranking '1' to the biggest number in a column, '2' to the second biggest value and so on. Obviously the smallest value in the column will get the lowest ranking. This should be done for both variables, after which the difference in ranks of the two values on each row of the table should be found and then squared. Note that the rank of the second value is subtracted from the rank of the first and that the differences are squared to remove negative values. Once these differences have been squared, the formula, evident on page 47, should then be used to find the correlation coefficient.
Hypothesis 1
A shopping centre which contains the greatest number of shops will have a large sphere of influence
For this hypothesis, it was first necessary to construct a desire line (figure 6) in order for the spheres of influences to be calculated for each shopping site. With Kingston Town Centre (orange line) clearly showing a bigger sphere of influence, this meant that more shoppers were willing to travel greater distances to get to Kingston. A scatter graph (page 26, graph 3) was then produced using the information in table 4 to check to see that shoppers were travelling from larger distances just to get to shopping centres with the greatest number of shops. From looking at the graph itself, it is clear that a positive correlation is visible as when there is an increase in one variable, there is an increase in the other however this trend does not always follow for this hypothesis, for example as Kingston Town Centre, with 572 shops, contains the largest sphere of influence (205 squares), we would expect the second largest shopping centre, Surbiton, with 169 shops, to have the next biggest sphere of influence (43 squares) but instead New Malden, the third largest shopping centre, shows a greater sphere of influence of 117 squares . None the less to support the weak positive correlation, I have calculated a Semi Average Regression line (in purple) on a second scatter graph (see graph 11 on the following page) to show the correlation.
The scatter graph shows Kingston (the point with the greatest number of shops and sphere of influence) as an obvious anomaly compared to the other shopping centres in the borough. As it is the 7th largest retailing store in the UK, it homes many department and specialist stores and thus it is at the top of the tier of shopping centres in the Kingston borough.
SCATTER GRAPH SHOWING NUMBER OF SHOPS AGAINST SPHERE OF INFLUENCE WITH A SEMI AVERAGE REGRESSION LINE HERE!- GRAPH 11
Hypothesis 2
The amount of time it takes to travel to a shopping centre with many shops will be greater than travelling to one with not so many shops.
To attempt to try and prove this hypothesis correct, I also drew up a scatter graph (graph 5) having converted the minutes to seconds. Again there is a weak positive correlation showing the number of shops in a shopping centre and the average travel time to get to the shopping centre. Therefore as the number of shops in the shopping centre increases, then the average travel time to get to the centre too increases. From graph 5, Kingston Town Centre, the largest shopping centre with 572 shops has the largest average travel time of 1521 seconds, which complies with the trend that the size of the shopping centre is directly proportional to the average travel time. The larger centres have a greater number of shops and a greater variety of shops and services, which attracts large amounts of people to the centre, from long distances, creating a large sphere of influence and hence a greater average travel time.
However, although New Malden is the third largest shopping centre, it has the second highest average travel time of 890 seconds, 230 seconds greater than Surbiton. Just as New Malden also had a larger sphere of influence than Surbiton in hypothesis 1, it now also has a greater average time. This can be explained as for New Malden, with a larger sphere of influence than Surbiton, people are travelling to New Malden from further away than too Surbiton so the average travel time is higher.
Kingston Road 1 and 2 do not fit this trend. Both shopping centres contain far fewer shops than Surbiton, yet they have managed to gain a greater travel time than Surbiton, the second largest shopping centre. Both Kingston Roads contain the greatest percentage of specialist types of stores, so it could be that shoppers are prepared to travel further for items such as leather goods and office furniture.
Like for hypothesis 1, I have calculated a Semi Average Regression line (in purple) on a second scatter graph (see graph 12 on the following page) to show the correlation.
Hypothesis 3
More shoppers will use public transport methods if travelling to a shopping centre with many shops
For this hypothesis I drew up two scatter graphs. The first one was done by using the bus maps and finding out the numbers of bus routes that passed through each shopping area. This scatter graph (graph 7) showed a rather weak positive correlation, and did demonstrate an expected trait, that there would be a higher number of bus routes present in shopping centres with many shops. The shopping centre with 572 shops, Kingston, has the highest number of bus routes, 24, which pass through that area.
The second scatter graph for this hypothesis (graph 10), has been constructed by using the information from table 2, as it shows the amount of people taking public transport and also the number of shops in each centre. There is a positive correlation evident from the scatter graph with most of the data focused towards the start of the scatter graph. Graph 10 successfully shows the correlation between number of shoppers using public transport and the number of shops in the centre as when there is an increase in one variable, there is an increase in the other. From this graph, it is evident that Kingston again appears as an anomaly due to the fact that it has more transport routes and shops than any other shopping centre in the Kingston borough. This is because Kingston is the 7th largest retailing store in the UK and also due to the various shops and products available to customers, more shoppers would be willing to travel from further distances and therefore there are more public transport routes.
Now that I have shown that my hypotheses show a positive correlation, I will now conduct a statistical test with my hypotheses. But for each hypothesis I will make a H1 and H0, where the H1 is the actual or alternative hypothesis and the H0 is the null hypothesis (i.e. states no link between two data sets in comparison to H1).
Hypothesis 1
H1= A shopping centre which contains the greatest number of shops will have a large sphere of influence.
H0= A shopping centre which contains the greatest number of shops has no affect on the sphere of influence.
Hypothesis 2
H1= The amount of time it takes to travel to a shopping centre with many shops will be greater than travelling to one with not so many shops.
H0= The amount of time it takes to travel to a shopping centre with many shops has no affect travelling to one with not so many shops.
Hypothesis 3
H1= More shoppers will use public transport methods if travelling to a shopping centre with many shops.
H0= Shoppers using public transport methods has no affect if travelling to a shopping centre with many shops.
For the statistical test, I will be using the Spearman Rank Correlation Coefficient used to discover the strength of a link between two sets of data. Now that I have produced 3 null hypotheses, the way in which a hypothesis should always be tested when conducting research, I now need to find whether I am to accept the H0 or the H1 I believe will occur.
To find the strength of the correlation, I must use the Spearman Rank formula:
n = number of pairs of values
d = difference in ranks
rs = Spearman rank coefficient
Spearman Rank Correlation Coefficient for Hypothesis 1- number of shops and sphere of influence
Correlation coefficient (r value) = 6∑ d² 6 x 12.5
1- = = 0.777
n³-n 73 - 7
Using the formula, I have been able to reject my H1 and accept my H0 to show that the statistical test was not successful and that my hypothesis has not been proven. This is because it has not reached the 5% significance level with the Spearman Rank correlation test. This is for the reason that the spheres of influences for Kingston Road 1 and Kingston Road 2, due likely to operator variants, were not measured and so instead of using ‘n’ (number of shops) as 9, I had to use it as 7. When plotting 7 on the ‘x’ axis on the Spearman Rank correlation graph, I had to find the degrees of freedom (7 – 2), which meant that the degrees of freedom was 5. With this being a lower figure for the Spearman rank graph, it was much harder to reach the 5% significance level, and so I must accept the null hypothesis instead.
If I was able to accept my H1 for this hypothesis, it would have been clear that this would be due to the fact that when there are more shops in a shopping centre, there are more products available for consumers to purchase and they also come in wide varieties. This would have meant that shoppers would travel further distances to be able to arrive at the shopping centre.
TABLE HERE FOR HYP 1
Spearman Rank Correlation Coefficient for Hypothesis 2- number of shops and average time taken
Correlation coefficient (r value) = 6∑ d² 6 x 28.5
1- = = 0.763
n³-n 93 - 9
By looking at where the point lies on the Spearman Rank correlation graph (see page 48), at 7 degrees of freedom, 0.763 is significant at the 1% level, giving me 99% confidence that there is significant correlation between the size of the shopping centre and the average travel time. I have sufficient confidence therefore, to reject H0 and to accept H1.
TABLE HERE FOR HYP 2
Spearman Rank Correlation Coefficient for Hypothesis 3- number of shops and number of bus routes
Correlation coefficient (r value) = 6∑ d² 6 x 14
1- = = 0.883
n³-n 93 - 9
Using the formula, I have again been able to reject my H0 and accept my H1 to show that the statistical test was successful and that my hypothesis has been proven. By looking at the Spearman rank graph on the following page, it is clear that the hypothesis is quite accurate and is just over the 1% significance level, proving that the hypothesis is accurate and there is a link between the two variables.
TABLE HERE FOR HYP 3
Spearman Rank Correlation Coefficient for Hypothesis 3- buses and trains (public transport)
Correlation coefficient (r value) = 6∑ d² 6 x 9
1- = = 0.925
n³-n 93 - 9
For the same hypothesis, I have once more been able to reject my H0 and accept my H1 to show that the statistical test was successful and that this doubles the proof for this hypothesis. The point on the Spearman Rank correlation graph (page 52) is over the 1% significance level and thus this hypothesis is correctly proven, to reveal that when there are more shops in a shopping centre there are also more public transport routes available for shoppers. Quite simply this is because when there are greater varieties in different products in shops which vary from convenience to department in shopping centres such as Kingston, then a greater proportion of customers would be attracted to the centre from local and further distances to purchase a good and they will take whatever means of transport to get there.
TABLE HERE FOR HYP 3
Conclusions- section 5
The purpose for carrying out this investigation was to initially investigate into the Royal Borough of Kingston upon Thames area shopping centres and their patterns of use. At the start of the investigation, I chose 3 hypotheses to try and establish a shopping hierarchy for all the shopping centres:
1A shopping centre which contains the greatest number of shops will have a large sphere of influence
This hypothesis was created to investigate into whether the size of a centre had an affect on the sphere of influence. The trends showed that as the size of the shopping centre increased so too did the sphere of influence. The larger shopping centres had the largest spheres of influences due to their wide range of shops and services attracting people to their centres for purchasing high order goods. The Spearman Rank coefficient also produced further evidence for the correlation to be positive between the two variables; however, there was not sufficient confidence for the rejection of the null hypothesis as the Spearman Rank correlation coefficients and degrees of freedom did not reach the 5% significance level. This was due to the little amount of data used for the hypothesis as not enough spheres of influence were found for enough shopping centres, so the degrees of freedom was even smaller than usually. Therefore, I had to reject this hypothesis.
2The amount of time it takes to travel to a shopping centre with many shops will be greater than travelling to one with not so many shops.
I can accept this hypothesis. The general trend was that as the shopping centre increased in size, the travel time to the centre increased. This is closely related to hypothesis 1, as shoppers are willing to travel a further distance for high order and specialised goods. The wider range of shops and services attracts people from further distances to the larger centres increasing the average travel time. The Spearman’s rank coefficient achieved the 5% significance level, providing sufficient confidence to reject the null hypothesis and accept the alternative.
3More shoppers will use public transport methods if travelling to a shopping centre with many shops.
This hypothesis was also very successful in proving a hierarchy amongst shopping centres in the Kingston borough. It is clear from the data that more people take the bus or rail to get to the larger shopping centres such as Kingston as there are more routes available to shoppers and the larger shopping centres are easily accessible so people travel from further distances.
From the data I have collected and interpreted though the usage of graphs and statistical tests that I have carried out, I can now make a hierarchy of shopping centres in the Kingston borough which will conclude my study.
I am able to create a hierarchical shopping system (figure 6) because the hypotheses I have investigated and tested shown through the presentation of the data, show that Kingston is the highest ranked shopping centre offering higher order goods in comparison to the centres that I have also investigated. Then also with this information, I am able to class the other centres in the two other tiers and a general hierarchy pattern emerges.
On the whole the investigation was conducted rather well and produced good credible results that were useful in proving the hypotheses. However improvements could have been made in this investigation if it were to be carried out again. These improvements could be:
- Making the investigation fairer by conducting the questionnaires on the same day and time. In doing this would provide representative results of the shopping.
- Carrying out the investigation on a weekend so that a better desire line map and spheres of influences could be produced as more people might tend to shop then.
- If there was not any operator variants, as it affected the investigation and resulted in some areas being anomalous and not used in the investigation.
Bibliography
Although I used my own knowledge and primary data for the investigation, it was necessary for me to also use other resources for information. The websites and books I used have been listed below.
- Microsoft Word- used for spelling and grammar check
- Microsoft Excel- used for drawing up graphs
- Tomorrow’s Geography for Edexcel GCSE specification A (Mike Harcourt & Stephen Warren)
- Collins Study and Revision Guide Geography GCSE (Michael Raw & Nicholas Rowles)
-
used for any maps which were relevant for the coursework.
-
used as a search engine for useful information such as Walter Christaller and his theories.
-
used again for vital information that was used in the coursework.
-
used for information on the Spearman Rank correlation coefficient.
-
used in order to obtain the bus maps for the number of bus stops in each shopping area to be counted.