Last updated on August 30th, 2019 at 07:00 am
The research question I wish to answer using bivariate correlation is: which of the continuous variables in the cycle touring dataset have a medium or strong correlation (with continuous and dichotomous variables)? For the definition of medium and high correlation, I will use the values defined by Cohen and specified in Pallant (2010, p. 134), specifically medium r = .30 to .49 and high r = .50 to 1.0 (absolute values). The continuous variables are: Age, PFitness, PWeight, Panniers, BudgetAccom, Budget, RideDays, TotalDist, TotalDays, and Variety. The dichotomous variables are: LocNA, Gender, TouringBike, DoAgain, and Blog. I used SPSS to generate the Pearson correlation table, excluding missing values pair-wise, and then edited the table to remove the bottom half and all the non-significant correlations, creating a table similar to the one in the SPSS Survival Manual (Pallant, 2010, p. 135). I found the simpler table structure easier to read and analyze, see Table 2. Table 2 shows the following medium correlations: Age and Budget (.331); Budget and BudgetAccom (-.422); RideDays and TotalDays (-.423); RideDays and TotalDist (-.381); and, RideDays and Variety (-.323). Table 2 shows the following high correlations: TotalDays and TotalDist (.914), TotalDays and Variety (.648), and TotalDist and Variety (.728). I then looked at the values for high and medium correlations to see if I could determine meaning out of the statistics. I was able to conclude that:
- As respondents get older their cycle touring budget increases.
- The more money respondents spend the less nights they stay in budget accommodation.
- For longer tours (in distance and time) and tours that visit more places, the respondents ride less days per week.
- The total distance, total days, and variety are highly correlated, such that a single factor may be present within this data.