CHAPTER 3: Research design and methodology
4.3 Multivariate analysis
4.3.2 Findings and analysis: Hypothesis 2
Table 4.8: Findings on hypothesis 2
Hypotheses Tools Findings
H2.1: The respondent‟s age will be
H2.5: Ethnicity will be related to cognitive evaluations.
H2.7: Property ownership will be related to cognitive evaluations.
The findings in Table 4.8 above are discussed in the following section.
4.3.2.1 Will age be related to cognitive evaluations?
H2.1: provided that the respondent‟s age will be related to cognitive evaluations. To test this hypothesis, an initial ANOVA test between age and cognitive evaluations was carried out. This test showed that there was a significant relationship between age and cognitive evaluations. Therefore, a Tukey HSD test was subsequently carried out to determine a more detailed relationship between age and cognitive evaluations. The results of the Tukey HSD test are shown in Table 4.9 below.
Table 4.9: Relationship between age and cognitive evaluations
Table 4.9 above shows that at a 5% level of significance different ages have significantly different cognitive evaluations of the city. The results show that the first four age groups have similar perceptions of the city cognitively. Thus, those aged 30 years and under cognitively perceive the city in a similar manner whilst the older respondents, that is 31 years and above perceive the city differently. In this case, older respondents had more positive cognitive evaluations than younger respondents. This is in line with the previous literature findings (Hidalgo and Hernández, 2001).
H2.1: The respondent‟s age will be related to cognitive evaluations.
{1} {2} {3} {4} {5} {6} {7} {8} {9}
< 18 Years 0.132938 0.833079 0.823376 1.000000 0.997242 0.992211 0.997556 0.647142 18-21 Years 0.132938 0.058030 0.053432 0.000010 0.000010 0.000010 0.000022 0.000010 22-25 Years 0.833079 0.058030 1.000000 0.001614 0.000010 0.000011 0.014314 0.000017 26-30 Years 0.823376 0.053432 1.000000 0.000876 0.000010 0.000011 0.012307 0.000015 31-40 Years 1.000000 0.000010 0.001614 0.000876 0.325456 0.311740 0.890465 0.040069 41-50 Years 0.997242 0.000010 0.000010 0.000010 0.325456 0.999996 1.000000 0.613066 51-60 Years 0.992211 0.000010 0.000011 0.000011 0.311740 0.999996 1.000000 0.794942 61-65 Years 0.997556 0.000022 0.014314 0.012307 0.890465 1.000000 1.000000 0.911399
> 66 Years 0.647142 0.000010 0.000017 0.000015 0.040069 0.613066 0.794942 0.911399 Age
Tukey HSD test; Variable: COGN_EVAL Marked differences are significant at p < .05000
4.3.2.2 Will gender be related to cognitive evaluations?
H2.2: provided that the respondent‟s gender will be related to cognitive evaluations. To test this hypothesis, which involved categorical data, T-tests were carried out on the male and the female responses. The test results showed that there was no significant relationship between gender and cognitive evaluations. However, T-tests on the other framework variables of envisioned future and affective evaluations showed that gender only has a significant relationship with the affective evaluations variable. Table 4.11 below shows the results of the T-tests between the categorical gender groups and the three framework variables.
Table 4.10: Relationship between gender and the framework variables
Table 4.10 above shows that at a 5% level of significance different genders do not have significantly different cognitive evaluations of the city. However, gender has significant relationship with affective evaluations. The relevant Cohen d‟s coefficient of 0.19 however shows that these gender differences although statistically significant, are of small practical significance. Therefore, it can be concluded that in this treatise, gender did not have an effect on city perceptions of the respondents. The small influence of gender on affective evaluations however, confirms the findings in the literature.
4.3.2.3 Will length of stay be related to cognitive evaluations?
H2.3: provided that the respondent‟s length of stay in the city will be related to positive cognitive evaluations. The current test thus seeks to find out if the amount of years that a person has lived in the city determines whether that person would have positive cognitive evaluations of the city. To test this hypothesis, an initial ANOVA test between age and the three framework
H2.2: Gender will be related to cognitive evaluations.
Variable Mean Mean t-value df p Valid N Valid N Std.Dev. Std.Dev. F-ratio p
COGN_EVAL 3.539380 3.574404 -1.48770 3657 0.136917 1742 1917 0.695468 0.725257 1.087501 0.073624 ENV_FUTURE 4.172304 4.146062 1.42617 3657 0.153903 1742 1917 0.554405 0.557234 1.010232 0.828513 AFF_EVAL 3.381107 3.232481 5.67724 3657 0.000000 1742 1917 0.764536 0.814081 d = 0.19 Small
T-tests; Group 1: Female; Group 2: Male
variables was carried out. This test showed that there was a significant relationship between length of stay and cognitive evaluations as well as envisioned future. A Tukey HSD test was subsequently carried out to determine a more detailed relationship between length of stay and cognitive evaluations. The results of the ANOVA and Tukey HSD test are shown in Table 4.11 and Table 4.12 below.
Table 4.11: Relationship between length of stay and the framework variables
Table 4.11 shows that length of stay is significantly related to cognitive evaluations as well as envisioned future.
Table 4.12: Relationship between length of stay and cognitive evaluations
Table 4.12 above shows that at a 5% level of significance respondents who have lived in the city the shortest time differ from all the other groups.
Respondents who have lived in the city the shortest have significantly lower cognitive evaluations of the city than the other respondents. Therefore, the longer the residents stay in the city, the higher their evaluations of the city‟s attributes. These findings confirm findings in the literature that found that longer residential time increases the conceptual perceptions of residents (Knez, 2005). However, the literature finding that increasing residential time increases emotional bonds was not confirmed in this treatise since there was no significant relationship between length of stay and affective evaluations.
H2.3: Length of stay will be related to cognitive evaluations.
SS df MS SS df MS F p
COGN_EVAL 25.66224 4 6.415559 1825.347 3654 0.499548 12.84274 0.000000 ENV_FUTURE 13.22141 4 3.305354 1117.467 3654 0.305820 10.80816 0.000000 AFF_EVAL 3.36068 4 0.840170 2304.227 3654 0.630604 1.33233 0.255371
Analysis of Variance
Marked effects are significant at p < .05000 Variable
H2.3: Length of stay will be related to cognitive evaluations.
{1} {2} {3} {4} {5}
< 5 Years 0.000075 0.000017 0.005787 0.480100
5-9 Years 0.000075 0.431746 0.176375 0.879103
10-20 Years 0.000017 0.431746 0.000021 0.391030
> 20 Years 0.005787 0.176375 0.000021 1.000000
N/A 0.480100 0.879103 0.391030 1.000000
Length of stay
Tukey HSD test; Variable: COGN_EVAL (Survey extractresults in Destination Identity variables.stw)
4.3.2.4 Will being born in a place be related to cognitive evaluations?
H2.4: provided that being born in the city will be related to positive cognitive evaluations. To test this hypothesis, which involved categorical data, T-tests were carried out on the “born in the city” and “not born in the city” responses.
The test results showed that there was no significant relationship between being born in the city and cognitive evaluations. However, T-tests on the other framework variables of envisioned future and affective evaluations showed that being born in the city only has a significant but small relationship with the envisioned future variable. Table 4.13 below shows the results of the T-tests between the born in the city and not born in the city categorical groups and the three framework variables.
Table 4.13: Relationship between being born in a city and framework variables
It can therefore be concluded that being born in a city will not influence a respondent‟s cognitive evaluations of the city. Interestingly, the literature speaks to a link between being born in a place and affective evaluations however; this treatise did not find a link between the two. Being born in the city did however influence a respondent‟s envisioned future of the city with the respondents who are born in the city having a slightly more positive evaluation of the city‟s future.
4.3.2.5 Will ethnicity influence cognitive evaluations?
H2.5: provided that ethnicity will influence cognitive evaluations. To test this hypothesis, an initial ANOVA test between ethnicity and cognitive evaluations was carried out. Due to the small number of respondents for the Asian This
H2.4: Being born in a place will be related to cognitive evaluations.
Mean Mean t-value df p Valid N Valid N Std.Dev. Std.Dev. F-ratio p
COGN_EVAL 3.571681 3.540893 1.303400 3657 0.192520 2001 1658 0.718679 0.702246 1.047348 0.325916 ENV_FUTURE 4.191289 4.119050 3.920206 3657 0.000090 2001 1658 0.543395 0.568432 d = 0.13 Small AFF_EVAL 3.315675 3.288232 1.040446 3657 0.298202 2001 1658 0.801894 0.784905 1.043758 0.363141 Variable
T-tests; Grouping: Q1-18 Were you born in Port Elizabeth? Group 1: Y; Group 2: N
carried out to determine a more detailed relationship between ethnicity and cognitive evaluations. The results of the Tukey HSD test are shown in Table 4.14 below.
Table 4.14: Relationship between ethnicity and cognitive evaluations
Table 4.14 above shows that at a 5% level of significance different ethnic groups do have significantly different cognitive evaluations of the city. In particular, blacks have more positive cognitive evaluations regarding the city than the other ethnic groups. Intriguingly, when ethnicity was tested against the other framework variables, it was found, as shown in Table 4.15 and Table 4.15 below, that Blacks also have higher envisions of the city‟s future but lower affective evaluations of the city. Whites on the other hand had similar affective evaluations to the other ethnic groups, except the Blacks, but had significantly lower envisioned future evaluations from all the other ethnic groups.
This means that blacks are more positive about the city their cognitive evaluations at the same time, they are more positive regarding the city‟s future however, they have a weaker emotional bond to the city than other groups. Whites on the other hand have the opposite evaluations. They rate the city lower in terms of cognitive evaluations as well as the city‟s future prospects. They however, have higher levels of emotional bonds with the city as shown in Table 4.16 below.
H2.5: Ethnicity will be related to cognitive evaluations.
Ethnicity
Black 0.000008 0.000008 0.000008
White 0.000008 0.158323 0.426912
Coloured 0.000008 0.158323 0.998347
Asian/Indian/Other 0.000008 0.426912 0.998347 Tukey HSD test; Variable: COGN_EVAL Marked differences are significant at p < .05000
Table 4.15: Relationship between ethnicity and envisioned future
Table 4.16: Relationship between ethnicity and affective evaluations
It can therefore be concluded that ethnicity does have a relationship with cognitive evaluations as well as with the other proposed framework variables.
4.3.2.6 Will income influence positive cognitive evaluations?
H2.6: provided that income will influence cognitive evaluations. To test this hypothesis, an initial ANOVA test between income and cognitive evaluations was carried out. This test showed a significant relationship. A Tukey HSD test was subsequently carried out to determine a more detailed relationship between income and cognitive evaluations. The results of the Tukey HSD test are shown in Table 4.16 below.
Table 4.17: Relationship between income and cognitive evaluations
H2.5: Ethnicity will be related to cognitive evaluations.
Ethnicity
Black 0.000467 0.868946 0.986451
White 0.000467 0.276565 0.581238
Coloured 0.868946 0.276565 0.999498
Asian/Indian/Other 0.986451 0.581238 0.999498 Tukey HSD test; Variable: ENV_FUT
Marked differences are significant at p < .05000
H2.5: Ethnicity will be related to cognitive evaluations.
Ethnicity
Black 0.000008 0.018300 0.000345
White 0.000008 0.725421 0.436909
Coloured 0.018300 0.725421 0.189494
Asian/Indian/Other 0.000345 0.436909 0.189494 Tukey HSD test; Variable: AFF_EVAL
Marked differences are significant at p < .05000
H2.6: Income will be related to cognitive evaluations.
Income
< R1000 0.366718 0.000451 0.000032 0.275040 0.839610 0.019439 R1000-R1999 0.366718 0.269611 0.018622 1.000000 0.000411 0.000026 R2000-R4999 0.000451 0.269611 0.838571 0.141321 0.000026 0.000026 R5000-R9999 0.000032 0.018622 0.838571 0.003835 0.000026 0.000026 R10000-R19000 0.275040 1.000000 0.141321 0.003835 0.000049 0.000026 R20000-R40000 0.839610 0.000411 0.000026 0.000026 0.000049 0.093859
>R40000 0.019439 0.000026 0.000026 0.000026 0.000026 0.093859 Tukey HSD test; Variable: COGN_EVAL (Survey extractresults in Destination Identity variables.stw)
Table 4.17 above shows that at a 5% level of significance different income groups do have significantly different cognitive evaluations of the city. In this regard, those who earn less than R1999 have a less positive evaluation of the city‟s cognitive evaluations. The next two income groups, earning between R2000 and R9999, has slightly more positive cognitive evaluations. Then the next two income groups earning between R10000 and R40000 have once again lower perceptions of the city cognitively. Then the highest earners have the most positive cognitive evaluations of the city.
This wavy trend could be explained by linking income groups with classes. In this case, the city seems to have a 4-tiered class structure. One could divide the outliers as the lower and upper classes and then divide the middle into a lower middle and upper middle class. In this case, the lower class has the lowest cognitive evaluations of the city followed by the upper middle class.
The lower middle class and the upper class have the highest levels of cognitive evaluations.
4.3.2.7 Will property tenure influence positive cognitive evaluations?
H2.7: provided that property tenure will influence cognitive evaluations. To test this hypothesis, an ANOVA test between property tenure and cognitive evaluations was carried out. This test showed that there was no significant relationship between property tenure and cognitive evaluations. However, further ANOVA tests on the other framework variables of envisioned future and affective evaluations showed that property tenure has a significant relationship with both envisioned future and affective evaluations variables.
Table 4.18 below shows the results of the ANOVA tests between property tenure and the three framework variables.
Table 4.18: Relationship between property tenure and framework variables
Table 4.18 above shows that at 5% level of significance different property tenure do not have significantly affect cognitive evaluations of the city.
However, it has a significant relationship with both envisioned future and affective evaluations. In this regard, subsequent Tukey HSD tests on the envisioned future and affective evaluations variables as highlighted in Table 4.19 below, show that the informal settlement dwellers have lower levels of envisioned future and affective evaluations.
Table 4.19: Relationship between property tenure, envisioned future and affective evaluations
It can therefore be concluded that although respondents‟ cognitive evaluations do not differ according to their property tenure; informal settlement dwellers have less positive envisioned future evaluations as well as lower affective evaluations of the city. This result shows that without a secure tenure, residents‟ levels of affect as well as their perception of a city‟s future are significantly lower than those of other residents.
4.3.2.8 Will being born in a place determine different cognitive associations of the city attributes?
H2.8 provided that the respondent‟s age would determine different associations of the city attributes (Q4-13 in the questionnaire). To test this hypothesis, the Spearman‟s Rank Correlation statistic was used.
H2.7: Property tenure will be related to cognitive evaluations.
Variable
COGN_EVAL 0.48091 2 0.24046 1850.528 3656 0.506162 0.47506 0.621886 ENV_FUTURE 5.89912 2 2.94956 1124.789 3656 0.307656 9.58721 0.000070 AFF_EVAL 24.45026 2 12.22513 2283.137 3656 0.624491 19.57616 0.000000
Analysis of Variance
Marked effects are significant at p < .05000
H2.7: Property tenure and cognitive evaluations H2.7: Property tenure and cognitive evals.
Property tenure Tukey HSD test; AFF_EVALProperty tenure
Rented 0.209357 0.003240 Rented 0.933529 0.000022
Owned 0.209357 0.000079 Owned 0.933529 0.000022
Informal Settlement 0.003240 0.000079 Informal Settlement0.000022 0.000022 Tukey HSD test; ENV_FUTURE
Marked differences are significant at p
The Spearman correlation ranking value measures the relationship between rankings of different ordinal variables or different rankings of the same variable, where a “ranking” is the assignment of the labels “first”, “second”,
“third” etc. (Diaconis, 1988). According to Diaconis (1988), the correlation coefficient falls within the interval [−1, 1] and an increasing rank coefficient implies increasing agreement between the rankings. The coefficient is 1 if the agreement between the two rankings is perfect (i.e. the two rankings are the same), it is 0 if the rankings are completely independent and −1 if the disagreement between the two rankings is perfect (i.e. one ranking is the reverse of the other). In this treatise the Pearson‟s correlation coefficients between the city role components were calculated as follows:
The number of YES responses for each respondent who was born in the city was counted based on all the Q4-13 attributes. The number of these YES responses was then used to rank the attributes. Subsequently, a correlation coefficient was calculated between the logical values (1 and 2) and the Yes responses. The correlation coefficient for these was found to be -6.11% which rejects H2.8.
It can thus be concluded that being born in a city will not result in respondents associating different attributes to the city compared to respondents who were not born in the city. This finding is different from the findings in the literature which found different evaluations based on whether the respondents were born or not born in the city (Jacobs and Buijs, 2011). However, this result is in line with the rest of the findings in this treatise. The city seems to be perceived in a similar manner by different residents regardless of their employment status or indeed whether they were born in the city or not. The city therefore has a uniform identity to different respondents. It is only the finding on length of stay that implies that the longer residents stay in the city, the better their perception of it. However, the underlying identity of the city does not change.
4.3.2.9 Conclusions
The foregoing statistical analysis supports the literature findings on the following demographic items‟ influence on cognitive evaluations: age, length of stay, ethnicity and income. On the other hand, the influence of gender, being born in city as well as property tenure was not proven. It can therefore be concluded that in Nelson Mandela Bay, respondent who are black;
respondents who are older and respondents who have lived longer in the city (but not necessarily born in the city) have more positive evaluations of the city.
Being born in the city has been found not to influence cognitive evaluations together with gender and property tenure. In fact, testing being born in a city with Q4.13, which related to city attributes most associated with the city, also showed that it does not influence the perception of those cognitive attributes.
Thus, it may be concluded that how residents cognitively view their city does not depend on whether they were born in the city or not.
In addition, residents associate the city with similar attributes regardless of whether they were born in the city or not. As mentioned above, the factors that do seem to make a difference in how the city is cognitively perceived by the residents are age, length of stay in the city, ethnicity and income.