Effect of Surface and Vision on Balance

Keywords: repeated measures, ordinal regression

Description

The balance of subjects were observed for two different surfaces and for restricted and unrestricted vision. Balance was assessed qualitatively on an ordinal 4-point scale based on observation by the experimenter. Subjects were expected to be better balanced (show less sway) when standing on the normal surface than on foam, and when their eyes were open rather than closed or when their vision was restricted by a dome.

Equal numbers of male and female subjects were chosen. For both males and females, ten older (more than 24 years old) and ten younger subjects were selected.

The data is available in two formats. The is in univariate or "strung out form" which is suitable for entry to Minitab or S-Plus and to most mixed model programs. The second is in repeated measures format which is suitable for SPSS and for most special purpose repeated measures programs.

Univariate format:


	Variable		Description

	Subject		1 to 40
	Sex		male or female
	Age		Age of subject in years
	Height		Height in cm
	Weight		Weight in kg
	Surface		normal or foam
	Vision		eyes open, eyes closed, or closed dome
	CTSIB		Qualitive measure of balance, 1 (stable) - 4 (unstable)

Repeated measures format:


	Variable		Description

	Subject		1 to 40
	Sex		male or female
	Age		Age of subject in years
	Height		Height in cm
	Weight		Weight in kg
	NO1		Balance measure on normal surface with eyes open, first replicate
	NO2		as above, second replicate
	NC1		Balance measure on normal surface with eyes closed, first replicate
	NC2		as above, second replicate
	ND1		Balance measure on normal surface with dome, first replicate
	ND2		as above, second replicate
	FO1		Balance measure on foam surface with eyes open, first replicate
	FO2		as above, second replicate
	FC1		Balance measure on foam surface with eyes closed, first replicate
	FC2		as above, second replicate
	FD1		Balance measure on foam surface with dome, first replicate
	FD2		as above, second replicate

Download

Data file (tab-delimited text, univariate format)
Data file (repeated measures format)

Source

Steele, R. (1998). Honours Thesis, Department of Physiotherapy, University of Queensland.

Analysis

The data is best analysed using ordinal methods, but normal analysis of variance gives a good approximation. We see strong effects for Surface and Vision, and an interesting interaction between Sex and Vision. If ordinal regression is used, the two 4's in the response will need to be recoded to 3's.

> aov.ctsib <- aov(CTSIB~Sex*Surface*Vision+Age+Height+Weight+Error(factor(Subject)))
> summary(aov.ctsib)
Error: factor(Subject)
          Df Sum of Sq  Mean Sq  F Value     Pr(F)
      Sex  1   4.21875 4.218750 5.023902 0.0314418
      Age  1   0.01731 0.017312 0.020616 0.8866523
   Height  1   1.43036 1.430357 1.703342 0.2003669
   Weight  1   3.69075 3.690748 4.395130 0.0433382
Residuals 35  29.39075 0.839736

Error: Within
                    Df Sum of Sq  Mean Sq  F Value     Pr(F)
           Surface   1  58.10208 58.10208 479.5757 0.0000000
            Vision   2  36.83750 18.41875 152.0287 0.0000000
       Sex:Surface   1   0.01875  0.01875   0.1548 0.6942197
        Sex:Vision   2   1.21250  0.60625   5.0040 0.0071071
    Surface:Vision   2   0.20417  0.10208   0.8426 0.4313002
Sex:Surface:Vision   2   0.61250  0.30625   2.5278 0.0810208
         Residuals 430  52.09583  0.12115