A factor analysis of the SSQ (Speech, Spatial, and Qualities of Hearing Scale)

Abstract

Objective: The speech, spatial, and qualities of hearing questionnaire (SSQ) is a self-report test of auditory disability. The 49 items ask how well a listener would do in many complex listening situations illustrative of real life. The scores on the items are often combined into the three main sections or into 10 pragmatic subscales. We report here a factor analysis of the SSQ that we conducted to further investigate its statistical properties and to determine its structure. Design: Statistical factor analysis of questionnaire data, using parallel analysis to determine the number of factors to retain, oblique rotation of factors, and a bootstrap method to estimate the confidence intervals. Study sample: 1220 people who have attended MRC IHR over the last decade. Results: We found three clear factors, essentially corresponding to the three main sections of the SSQ. They are termed “speech understanding”, “spatial perception”, and “clarity, separation, and identification”. Thirty-five of the SSQ questions were included in the three factors. There was partial evidence for a fourth factor, “effort and concentration”, representing two more questions. Conclusions: These results aid in the interpretation and application of the SSQ and indicate potential methods for generating average scores.

Key Words::

• Psychoacoustics/hearing science
• behavioral measures
• hearing aids
• speech perception

Notes

Acknowledgements

We thank the staff of the Scottish Section for collecting the data over the past 10 years, Graham Naylor (Eriksholm) & Oliver Zobay (IHR) for comments on the manuscript, Kelly Demeester (Leuven) for her cluster-analysis data, and the editors and reviewers of IJA for their informative comments during submission. Earlier versions of the analysis were presented at the 2011 ICRA meeting and the 2011 British Society of Audiology Annual Conference.

Declaration of interest: The authors report no declarations of interest. This work was supported by the Medical Research Council (grant number U135097131) and by the Chief Scientist Office of the Scottish Government.

Notes

1. Gatehouse and Noble (2004) list 50 items. One item, however, asks specifically about the effect of hearing aids or cochlear implants (Qualities #15). It is therefore excluded from many studies that are concerned with unaided listening.

2. We encourage users of the SSQ to ensure that all questions are answered.

3. The 49 questions used by Noble et al (2013) are the 48 questions considered here, plus Qualities #17.

4. This description of the dependence of SSQ scores on hearing loss, etc. is deliberately limited, as the present paper concentrates on the factor analysis itself. We hope to publish other analyses in the future.

5. As we earlier excluded those people who did not fully complete the SSQ, none of these matrices had any missing observations.

6. Occasionally we came across what is known as the alignment problem: the factors were reported in a different order or with negative weights instead of positive (Zhang et al, 2010). We silently corrected all of these by hand.

7. Many bootstrap applications use thousands of replications in order to obtain very accurate confidence intervals, but Efron and Tibshirani (1994, p. 52) suggested that 50 replications was often enough to get a good estimate of the standard error. Also we did not apply either the bias or acceleration corrections, mainly to simplify the calculations (Efron & Tibshirani, 1994; see Zhang et al, 2010 for their effects in bootstrapped factor analyses). Thus our estimates of the confidence intervals can only be estimates, but we contend that they are sufficiently accurate for present purposes.