Advertisement
Research Article

An Investigation of Racing Performance and Whip Use by Jockeys in Thoroughbred Races

  • David Evans,

    Affiliation: Faculty of Veterinary Science, University of Sydney, Sydney, New South Wales, Australia

    X
  • Paul McGreevy mail

    paul.mcgreevy@sydney.edu.au

    Affiliation: Faculty of Veterinary Science, University of Sydney, Sydney, New South Wales, Australia

    X
  • Published: January 27, 2011
  • DOI: 10.1371/journal.pone.0015622

Reader Comments (4)

Post a new comment on this article

Concerns re interpretation

Posted by nigel on 23 Mar 2011 at 00:57 GMT

I read with interest a recent paper titled An investigation of racing performance and whip use by jockeys in Thoroughbred races by Evans and McGreevy (2010) that assessed statistical associations between measures of racing performance and whip use in Australian racehorses. The paper has stimulated a good deal of interest and debate about this topic. I wish to raise concerns over a number of issues related to this paper. My motivation in doing this is to ensure that there is rigorous science-based critique of research that has the potential to provide input into debate over issues such as racing policy.

The paper describes the use of integer measures such as counts of whip strikes and race placing (first = 1 and so on) as well as sectional times (measured in seconds) in both linear regression and logistic regression models. Regression models are generally associated with assumptions about independence of observations and particularly about distributional patterns of model residuals (see Dohoo et al 2010 for more detailed discussion). The authors have not provided any information about model checking that may have been implemented for each of the regression models. Provision of such information would help provide confidence in the statistical outputs of the various models described in the paper.

The authors provide evidence of a statistical association between use of the whip in the final 200m of a race and the odds of a horse finishing in the first three places. This is dismissed as being non-causal in the discussion with little explanation other than reference to another finding of the same paper, that average sectional times in the final 600m of a race, as measured over successive 200m sections, was shown to be slowing as horses finished the race.

I would like to suggest that the association between whip use and race outcome (odds of placing in the first, second or third position) as demonstrated in the results of the paper, may indeed be causal and that such causality is not necessarily dependent on demonstration of faster sectional times at the end of a race by those horses that may place in the first three positions.

It seems plausible that whip strikes may influence a horse’s behavior and actions in a meaningful way without necessarily influencing speed or sectional time when measured over a 200m length of racetrack. Whip strikes in conjunction with a range of jockey actions (body position, use of reins, hands and legs) may encourage effort over short distances to take advantage of a gap in the field, cover an attempt by a competing horse to improve its position, maintain a particular course, and also encourage maximal effort over the finishing stages of a race without necessarily resulting in an increase in speed measured over a 200m distance.

It is also considered plausible that the average sectional time at the end of a race may indeed be slower than the middle sectional times while still potentially allowing individual horses in the race to either maintain sectional times (maintaining speed) near the end of the race or even reduce sectional time (increase speed). It is also plausible that horses finishing in the first three positions may not have the fastest sectional time in the final 200m of a race.

Average sectional time across all horses in a race may be influenced disproportionately by horses that are completely out of contention and that slow dramatically as they approach the finish line. Individual horses may speed up near the end of a race, have a faster final sectional time and yet still not place in the first three, if they were relatively slower in the first parts of a race. A horse may start very quickly in a race and build a positional advantage that they can hold onto at the end of a race even if they are travelling at a slower speed relative to competing horses in the final section. While accepting that absolute speed is important for race records, in many races relative speed is likely to be more important, meaning that the winning horse only has to be relatively faster than other horses in the same race.

The authors discuss a range of limitations associated with the study but do not mention the relatively small sample size and the relatively low r-squared value for statistical models, indicating that the statistical models provide relatively poor explanation of the variability in the outcomes being assessed. In conjunction with other limitations as mentioned (non-representative samples of races from a single racetrack), the effect is to suggest caution to avoid over interpretation of the findings of the paper.

The authors are applauded for providing scientific scrutiny of the use of the whip in racehorses. In my view this paper provides some useful and interesting observations on horse performance and whip use. It does not provide unequivocal support for a conclusion that whip use is not causally associated with performance. In some parts of the discussion the authors appear to be presenting an ideological position that may go beyond the methodology or the results of the paper and that may be more appropriate in a different forum.

I support the call by the authors for further and more detailed research on whip use in racehorses as a means of providing high-quality science-based information to policy makers. Further work is needed to continue the initiative demonstrated by these authors and provide valid and credible information to racing stakeholders.

Dohoo I, Martin W and Stryhn H (2009). Veterinary Epidemiologic Research 2nd edition. Published by VER Prince Edward Island, Canada 865 P.

Evans D and McGreevy P (2011). An investigation of racing performance and whip use by jockeys in Thoroughbred races. PLOS One, 6(1):e15622
Available at http://www.plosone.org/ar...


Competing interests declared: I am the Manager of the Horse Research Program for the Rural Industries Research and Development Corporation (RIRDC) in Australia (www.rirdc.gov.au)

RE: Concerns re interpretation

PaulMcG replied to nigel on 01 Apr 2011 at 02:02 GMT

Comment by “nigel” (1)

I read with interest a recent paper titled An investigation of racing performance and whip use by jockeys in Thoroughbred races by Evans and McGreevy (2010) that assessed statistical associations between measures of racing performance and whip use in Australian racehorses. The paper has stimulated a good deal of interest and debate about this topic. I wish to raise concerns over a number of issues related to this paper. My motivation in doing this is to ensure that there is rigorous science-based critique of research that has the potential to provide input into debate over issues such as racing policy.

Authors’ responses (1)

We appreciate an opportunity to address this feedback.

Comment by “nigel” (2)

The paper describes the use of integer measures such as counts of whip strikes and race placing (first = 1 and so on) as well as sectional times (measured in seconds) in both linear regression and logistic regression models. Regression models are generally associated with assumptions about independence of observations and particularly about distributional patterns of model residuals (see Dohoo et al 2010 for more detailed discussion). The authors have not provided any information about model checking that may have been implemented for each of the regression models. Provision of such information would help provide confidence in the statistical outputs of the various models described in the paper.

Authors’ responses (2)

The statistics methods are described in detail. If other scientists have sufficient concerns with any aspects of the methodology, the study should be independently repeated.

It would have been useful for the critique to be specific about the limitations of the model used, rather than simply call the used model into question because the limits or otherwise of alternative models have not been discussed in the paper.

Comment by “nigel” (3)

The authors provide evidence of a statistical association between use of the whip in the final 200m of a race and the odds of a horse finishing in the first three places. This is dismissed as being non-causal in the discussion with little explanation other than reference to another finding of the same paper, that average sectional times in the final 600m of a race, as measured over successive 200m sections, was shown to be slowing as horses finished the race.

Authors’ responses (3)

An association between “use of the whip in the final 200m of a race and the odds of a horse finishing in the first three places” was significant only when placings at the 400 and 200 m positions were excluded from the model. To interpret that result as meaningful, and potential cause and effect, in the absence of consideration of placings at 400 and 200 would be errant, because those placings were themselves associated with number of future whip strikes. Performance in the first 850-1050 metres predicted whip use and likelihood of finishing in the first 3 placings!

We did show an association between such whip use and final position (1 to n, where n=number of starters). Of course, that association has little “real-world” importance, because it is not too important whether or not a horses finishes 6th compared to finishing 7th in a race with 10 starters.

Imagine the outrage if we concluded that the association between whip use and final placing (1 to n) or finishing in the first 3 places was cause and effect, having ignored the association between whip use and prior performance to the 400 and 200 m positions.

In no part of the paper do we link slowing of the horses (or absence of acceleration) as a rationale for dismissing a possible cause and effect linkage between whip use and final placing. The slowing was presented only as a descriptive statistic.
Velocity was not used as an outcome variable.

If whipping more frequently was truly causative of superior performance, the act of whipping must have impacted velocity. We argued that the frequency of whipping did not influence likelihood of finishing in the first 3 places, after consideration of their placings at the 400 and 200. We also clearly stated ““.....it remains possible that whip use in the final stages of a race really does improve relative performance at a stage when all horses are slowing, but more frequent and sensitive methods of measuring velocity may be required to detect such a cause and effect linkage.”

As well, ‘The results in this study do not support a conclusion that whipping cannot affect velocity of an individual Thoroughbred racehorse during the final 400m section of a race”
As well, “Any effect of whipping on velocity in this study may have been a transitory. Such an effect, if it exists, may differ among horses”

Comment by “nigel” (4)

I would like to suggest that the association between whip use and race outcome (odds of placing in the first, second or third position) as demonstrated in the results of the paper, may indeed be causal and that such causality is not necessarily dependent on demonstration of faster sectional times at the end of a race by those horses that may place in the first three positions.

Authors’ responses (4)

See response above.

We in no part of this paper argued that causality of an association between whip use and race outcome would have to be contingent on whether horses were accelerating or showing reduced deceleration over 200 m distances in the last 400 metres. We clearly pointed out the limitations of measuring velocity every 200 metres.

Comment by “nigel” (5)

It seems plausible that whip strikes may influence a horse’s activity in a meaningful way without necessarily influencing speed or sectional time when measured over a 200m length of racetrack.

Authors’ responses (5)

Yes. See above.

In no part of the report was it argued that responses to whip strikes depended on a demonstration of an effect on 200 m sectional times. The limitations inherent in measuring velocity with sectional times every 200 m are clearly noted in the Discussion, and we suggested alternative approaches for future research to address this issue.

It is difficult to understand how it could be argued that more frequent shifting of a jockey’s body weight plus arm movements could positively influence performance, independent of whip strikes. What would be the mechanism? If anything, it is more likely that such jockey motion would be more likely to negatively impact economy of locomotion and so compromise the likelihood of superior performance. We look forward to a testing of that hypothesis.

Comment by “nigel” (6)

Whip strikes in conjunction with a range of jockey actions (body position, use of reins, hands and legs) may encourage effort over short distances to take advantage of a gap in the field, cover an attempt by a competing horse to improve its position, maintain a particular course, and also encourage maximal effort over the finishing stages of a race without necessarily resulting in an increase in speed measured over a 200m distance.

Authors’ responses (6)

How can responses over short distances such as “to take advantage of a gap in the field, cover an attempt by a competing horse to improve its position, maintain a particular course, and also encourage maximal effort over the finishing stages of a race” not impact on velocity? Can fatigued horses make such locomotory responses in a manner that affects the probability of finishing in the first 3 placings?

If there were such brief responses in conjunction with more whipping, such horses would presumably be more likely to finish in the first 3 placings, after taking into account their already superior relative placings before such jockey contributions of course.

Stewards confirmed that in this study the whip use was not associated with a need to correct steering in any horse.

In the Discussion, we noted the possibility that whip use might result in brief locomotory responses, and that such responses might vary between horses. We also suggest that future studies need to look at possible locomotory responses to whipping during and in the absence of fatigue, with measurements taken more frequently than in the PLoS paper.

Comment by “nigel” (7)

It is also considered plausible that the average sectional time at the end of a race may indeed be slower than the middle sectional times while still potentially allowing individual horses in the race to either maintain sectional times (maintaining speed) near the end of the race or even reduce sectional time (increase speed). It is also plausible that horses finishing in the first three positions may not have the fastest sectional time in the final 200m of a race.

Authors’ responses (7)

The critique is still assuming that we don’t understand this, and that we based our analyses and Discussion on an assumption that an effect of whipping must be associated with an effect on 200 m sectional times. At no point did we use 200m sectional times as an outcome variable in the logistic regression analyses.

Again, in the Discussion, we stated “.....it remains possible that whip use in the final stages of a race really does improve relative performance at a stage when all horses are slowing, but more frequent and sensitive methods of measuring velocity may be required to detect such a cause and effect linkage.”

As well, ‘The results in this study do not support a conclusion that whipping cannot affect velocity of an individual Thoroughbred racehorse during the final 400m section of a race”
As well, “Any effect of whipping on velocity in this study may have been a transitory. Such an effect, if it exists, may differ among horses”


Comment by “nigel” (8)

Average sectional time across all horses in a race may be influenced disproportionately by horses that are completely out of contention and that slow dramatically as they approach the finish line. Individual horses may speed up near the end of a race, have a faster final sectional time and yet still not place in the first three, if they were relatively slower in the first parts of a race. A horse may start very quickly in a race and build a positional advantage that they can hold onto at the end of a race even if they are travelling at a slower speed relative to competing horses in the final section. While accepting that absolute speed is important for race records, in many races relative speed is likely to be more important, meaning that the winning horse only has to be relatively faster than other horses in the same race.

Authors’ responses (8)

Horses “completely out of contention” are not allowed to be whipped.

On average, in this study, there was no slowing “as they approached the finish line”. Velocity was already significantly reduced at the 200 m mark.

It was also surprising to find in this study that horses in less advanced positions at the 400 m position were subsequently whipped less frequently over the last 400 metres. A decision was made by jockeys on horses with inferior placings, but still in contention, to whip less frequently at the 400 m mark, well before the finish. Why would jockeys do that if they believed that more whipping improves relative or absolute performance? It would be interesting to ask them that question.
The critique is still assuming that we don’t understand the myriad factors influencing race outcome, and that we based our analyses and discussion on an assumption that an effect of whipping must be associated with an effect on 200 m sectional times.

The aim for jockeys is to win, or finish in the first three positions. We understand that prize money and betting payments are not made to horses with the fastest final sectional times. That is why we were not interested in sectional times as an outcome variable, and we instead focussed on placings 1-3.
Win, second and third are relative positions. Approximately 95% of prize money and all of the betting is linked to finishing in the first three positions. That is why we used finishing in the first 3 or not, rather than acceleration or reduced deceleration over 200 metres, as the outcome variable.

Comment by “nigel” (9)

The authors discuss a range of limitations associated with the study but do not mention the relatively small sample size and the relatively low r-squared value for statistical models, indicating that the statistical models provide relatively poor explanation of the variability in the outcomes being assessed. In conjunction with other limitations as mentioned (non-representative samples of races from a single racetrack), the effect is to suggest caution to avoid over interpretation of the findings of the paper.

Authors’ responses (9)

Yes, we did not explicitly state that a larger sample size would be desirable in future studies which were designed to exactly replicate this study, with its defined limits on race length, race track, and ground conditions. Such a study would be of questionable merit. In the discussion we focussed on what we believe are more important follow-up research questions.

The Discussion notes the limitations of the sample size and selection criteria in the Discussion “Results of similar investigation in other racing circumstances, and in races with different prize money, might differ.”

Larger numbers of horses would clearly be needed in order to study horses in “other racing circumstances”. There would be little value in a further larger study with the same methodology (for measuring velocity) and same criteria for race selection (track, distance, track condition), given the acknowledged methodological limitations, e.g., frequency of velocity measurement.

Which causation was suggested? We did not argue any causality of these associations with low r squared values, except for the associations between superior positions at the 400 and 200 metre positions, and the outcome of superior performance at the finish.

The Discussion canvassed the view that an investigation of locomotory responses to whip use during races was a higher priority for future research than exact repetition of this study with larger numbers of horses. If the results from a study with larger numbers in an exact replicate of this study were the same as in this study, no progress would have been made. We are still left with the questions of whether or not there are locomotory responses to whip use, and how these responses vary between horses when struck while fatigued, and how race outcomes do or do not interact with whip use under different race circumstances.

See Discussion for “Highly sensitive, accurate and frequent measurements of velocity and/or position during such whipping in a large number of horses during races could address this issue.”

Comment by “nigel” (10)

The authors are applauded for providing scientific scrutiny of the use of the whip in racehorses. In my view this paper provides some useful and interesting observations on horse performance and whip use. It does not provide unequivocal support for a conclusion that whip use is not causally associated with performance. In some parts of the discussion the authors appear to be presenting an ideological position that may go beyond the methodology or the results of the paper and that may be more appropriate in a different forum.

Authors’ responses (10)

Where did the manuscript claim that the results “provide unequivocal support for a conclusion that whip use is not causally associated with performance”?

Our ideology was not the basis for the study. RSPCA Australia has an expressed view and ideology on this issue, but such ideology was absent in the design and rationale for the study, and in the analysis of the results. We had no idea what to expect in the Results of the investigation.

We agree that this paper does not answer every question that is needed to form “unequivocal” viewpoints. The results may contribute to changes in ideology, and may or may not inform revisions to administration of the whip rules in thoroughbred racing. Hopefully, they will also encourage administrators to continue to cooperate with researchers in future investigations.
Certainly, the conditions described in this study would be difficult to study in a controlled experiment. We are confident that an institutional ethics committee would disallow conduct of research that had these outcomes, i.e., that after having demonstrated onset of fatigue over a 12 second period, the exercise and the whipping stimulus would continue to be applied. Not only that, but it would be then applied more frequently, and in an unregulated manner for 6 of the final 12 seconds.

It would have been interesting to read of the details of the sentence(s) that the author of the critique has had difficulty with. It seems as though an ideology is being expressed in that feedback. Readers are, of course, entitled to disagree with any opinions, meanings or extrapolations expressed by the authors, and we look forward to future studies based on hypotheses arising from their disagreements or questions.

Comment by “nigel” (11)

I support the call by the authors for further and more detailed research on whip use in racehorses as a means of providing high-quality science-based information to policy makers.

Further work is needed to continue the initiative demonstrated by these authors and provide valid and credible information to racing stakeholders.

Authors’ responses (11)

The published paper was not a written as a report to racing stakeholders. Racing stakeholders have used the paper to justify their current policies.

Credibility in science depends on repetition and refinement of scientific investigations. Of course further work is required, given that this is the first study of interactions between whip use and race performance in Thoroughbred races. To expect that one study will provide “unequivocal” contributions to the issue denies the meaning and process of science, and the comment in the critique is an invalid contribution and seems to reflect a bias.

Potential conflicts of interests also need to be noted to help assure confidence in the validity and credibility of all science.

The published paper lists potential conflicts of interest. The critique should do the same.

In that context, the authors sought an answer to the following question from the author of the critique.

“Has any organisation requested a review of this paper by yourself or Ausvet?”

At the time of posting this response that question had not been answered.




Competing interests declared: I am one of the co-authors of the original article