Skip to main content

Advertisement

SpringerLink
Log in

Saving lives: the 2006 expansion of daylight saving in Indiana

  • Original Paper
  • Published:
Journal of Population Economics Aims and scope Submit manuscript

Abstract

Using data provided by the Indiana State Department of Vital Statistics, this study examines the mortality effects of daylight saving time observance using the April 2006 expansion of daylight saving time in Indiana as a natural experiment. The expansion of daylight saving time to all Indiana counties lowered the average mortality rate in the treatment counties during the months in which daylight saving time was observed. Stratified demographic analyses indicate that daylight saving time reduced mortality among males, females, and whites, but only among those aged 65 years and older. Specific-cause analysis indicates that daylight saving time lowered mortality primarily via reduced cancer mortality. The results of this study suggest a novel solar UVB-vitamin D mechanism could be responsible for the reduction in treatment county mortality following the expansion of daylight saving time in Indiana.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4

Similar content being viewed by others

Notes

  1. Farmers oppose DST because it means an extra hour of morning work in darkness immediately following the spring transition (Prerau 2005).

  2. Europe has a shorter DST observance period, switching to DST in late March and back to standard time in late October.

  3. With the exception of cold water ocean fish, food is a poor source of vitamin D: Three ounces of herring provides 1300 IU of vitamin D, 3.5 ounces of salmon provides 350 IU, and 1 teaspoon of cod liver oil provides approximately 400 IU. More commonly consumed fortified foods fail to provide sufficient daily vitamin D in typically consumed quantities, however. These foods include “vitamin D” milk or orange juice (100 IU per cup) and fortified breakfast cereals (50–100 IU per cup).

  4. Vitamin D insufficiency is determined by measuring blood concentration of 25-hydroxy-vitamin D3. Although there is lingering disagreement as to the definition of “deficiency,” as of 2011, the Institute of Medicine considers 25-hydroxy-vitamin D3 (or 25(OH)D3) concentrations less than 10 ng/mL (or 25 nmol/L) to be “severely deficient,” less than 20 ng/mL (or 50 nmol/L) to be “deficient,” and less than 30 ng/mL (or 75 nmol/L) to be “insufficient” (Institute of Medicine 2011).

  5. The duration of Vitamin D winter in Indiana was calculated using the free Android and IOS “dminder” Vitamin D tracker app available here: http://dminder.ontometrics.com/

  6. Available hours of UVB in northern Indiana and vitamin D3 synthesis estimates were calculated using the free Android and IOS “dminder” Vitamin D tracker app available here: See http://dminder.ontometrics.com/

  7. https://ods.od.nih.gov/factsheets/VitaminD-Consumer/

  8. Gibson, Jasper, Lake, La Porte, Newton, Porter, Posey, Spencer, Vanderburgh, and Warrick counties, all near the Illinois border, were mandated by the US Department of Transportation to observe DST as members of the Central time zone.

  9. Eight counties in western Indiana were also shifted from the Eastern to the Central time zone on April 2, 2006. These counties were Daviess, Dubois, Knox, Martin, Perry, Pike, Pulaski, and Starke. This resulted in 74 Indiana counties in the Eastern time zone and 18 in the Central. On March 11, 2007, Pulaski county was moved back to the Eastern time zone. On November 4, 2007, Daviess, Dubois, Knox, Martin, and Pike moved from Central to Eastern time. Ultimately, 80 Indiana counties now reside in the Eastern, and 12 remain in the Central time zone.

  10. My identification strategy is an extended version of the strategy used by Kotchen and Grant (2011).

  11. The daylight saving period was observed: April 6, 2003 through October 26, 2003; April 4, 2004 through October 31, 2004; April 3, 2005 through October 30, 2005; April 2, 2006 through October 29, 2006; March 11, 2007, through November 4, 2007; March 9, 2008, through November 2, 2008. Since daily mortality data is unavailable, in this study DSTt = 1 for April through October 2003–2006 and March through October 2007–2008.

  12. Bertrand et al. (2004) note the possibility of auto-correlation bias associated with the linear difference-in-difference estimator. I address this possibility in Appendix A, consistent with section IV.C of their paper, by collapsing the time series information into pre- and post-treatment time periods as a robustness test.

  13. The neighbors treatment group is composed of Allen, Benton, Crawford, Dearborn, Dubois, Elkhart, Jefferson, Knox, La Porte, Marshall, Noble, Pike, Pulaski, Ripley, Scott, Spencer, Starke, Steuben, Switzerland, Washington, and White counties.

  14. The Indiana State Department of Health follows a “Rule of Twenty” when calculating monthly county mortality rates. When the death count is less than 20, the mortality rate is unstable (i.e., the relative standard error rises exponentially with decreasing death count), meaning that a small change in the numerator can lead to a large change in the mortality rate from one month to the next. Mortality rates with relative standard errors greater than 25% are not considered reliable for estimation purposes. See https://www.stats.indiana.edu/vitals/CalculatingARate.pdf for more detail on their methods.

  15. In no month during 2003–2008 were there 20+ deaths within either the Pacific Islander or Asian demographic.

  16. https://www.census.gov/data/tables/time-series/demo/popest/intercensal-national.html

  17. DST treatment effects using raw death counts as dependent variables, weighted by population, generate corroborating estimates for the effect of DST on Indiana mortality. These results are available upon request.

  18. Consistent with the cyclic behavior of seasonal mortality and vitamin D3 levels, the adoption of DST by the treatment counties alone had no measurable effect on aggregate yearly mortality compared to any control group.

  19. Due to the sparseness of the county-monthly mortality data, particularly among demographic subgroups and narrowly defined specific causes, the full sample is used to estimate within-demographic and cause-specific treatment effects.

  20. The regression equation and the interpretation of the treatment effect estimators remain unchanged when using GLM with a log link function.

  21. Poisson regression requires the strict assumption that the variance equals the mean. I replicate the Poisson results in Appendix A using Poisson quasi-maximum likelihood estimation to guard against potential violations of this assumption and include the dispersion parameters to support using the Poisson model to estimate specific-cause DST treatment effects.

  22. The three largest causes of death in Indiana (all-cancers, all-cardiovascular disease, and all-respiratory disease) generate well-behaved log-normal distributions. As a robustness check, log-linear regression DST treatment effect estimates on these three mortality rates are comparable in magnitude and significance to those found using Poisson regression. These three results are available upon request.

  23. Bone cancer mortality lacked sufficient number of deaths to generate treatment effects. Ovarian cancer lacked sufficient number of deaths to generate treatment effects when examined separately from female reproductive cancers.

  24. April of 2003–2006 and March of 2007 and 2008.

  25. When examined in aggregate or individually, the DST period treatment effects on external cause mortality rates, including the motor vehicle accident mortality rate, are insignificant. This is likely because the external cause treatment effects are too small to significantly alter mortality rates at the Indiana county level: Smith (2016) reports that the spring DST transition is responsible for 30 more vehicular accident deaths in the entire USA. Similarly, Toro et al. (2015) estimate that DST is responsible for 196 extra heart attack deaths per year in all of Brazil (individual external causes of death listed in the ICD-10 include homicide, suicide, vehicular accident, other transport and non-transport accidents, falls, and medical complications)

  26. For example, Starke (Central time) and Pulaski (Eastern time) counties in northwest Indiana share a longitudinal meridian. See Fig. 4.

  27. Life savings is calculated as follows: DST was observed, on average, 7.67 months per year during the post-treatment period. The mortality decrease per year is then calculated: [2.99, 3.93]deaths/(month×100, 000population)×6.25M2006Indianapopulation×7.67monthsofDSTperyear = [1433.33, 1883.94]deathsperyear

  28. Although it is impossible to know for certain, a priori there is no example of a place that has sprung forward to DST without falling back to standard time.

  29. The synthetic control method fails to generate a comparable control without time series matching variables.

  30. In log linear regression, the variance is assumed proportional to the dependent variable.

References

  • Abadie A, Diamond A, Hainmueller J (2010) Synth: an R package for synthetic control methods in comparative case studies. J Stat Softw 42 (i13):493–505

    Google Scholar 

  • Abulmeaty MMA (2017) Sunlight exposure vs. vitamin d supplementation on bone homeostasis of vitamin d deficient rats. Clin Nutrit Exp 11:1–9

    Article  Google Scholar 

  • Al Mheid I, Quyyumi AA (2017) Vitamin d and cardiovascular disease. J Am Coll Cardiol 70(1):89–100

    Article  Google Scholar 

  • Autier P, Gandini S (2007) Vitamin d supplementation and total mortality: a meta-analysis of randomized controlled trials. Arch Int Med 167 (16):1730–1737

    Article  Google Scholar 

  • Balion C, Griffith L, Strifler L, Henderson M, Patterson C, Heckman G, Llewellyn D, Raina P (2012) Vitamin d, cognition, and dementia a systematic review and meta-analysis. Neurology 79:1397–405

    Article  Google Scholar 

  • Barbarawi M, Kheiri B, Zayed Y, Barbarawi O, Dhillon H, Swaid B, Yelangi A, Sundus S, Bachuwa G, Alkotob ML, Manson JE (2019) Vitamin D supplementation and cardiovascular disease risks in more than 83000 individuals in 21 randomized clinical trials: a meta-analysis. JAMA Cardiol 4(8):765–776

    Article  Google Scholar 

  • Barnes CM, Wagner DT (2009) Changing to daylight saving time cuts into sleep and increases workplace injuries. J Appl Psychol 94(5):1305

    Article  Google Scholar 

  • Benjamini Y, Hochberg Y (1995) Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Royal Stat Soc Series B (Methodological) 57(1):289–300

    Google Scholar 

  • Berk M, Dodd S, Hallam K, Berk L, Gleeson J, Henry M (2008) Small shifts in diurnal rhythms are associated with an increase in suicide: the effect of daylight saving. Sleep Biol Rhyt 6(1):22–25

    Article  Google Scholar 

  • Bertrand M, Duflo E, Mullainathan S (2004) How much should we trust differences-in-differences estimates? Quart J Econ 119(1):249–275

    Article  Google Scholar 

  • Bjelakovic G, Nikolova D, Bjelakovic M, Gluud C (2017) Vitamin D supplementation for chronic liver diseases in adults. Cochrane Datab Syst Rev 11:1–66

    Google Scholar 

  • Boscoe F, Schymura M, Boscoe FP, Schymura MJ (2006) Solar ultraviolet-B exposure and cancer incidence and mortality in the United States, 1993-2000. BMC Cancer 6:264

    Article  Google Scholar 

  • Chandler PD, Chen WY, Ajala ON, Hazra A, Cook N, Bubes V, Lee I-M, Giovannucci EL, Willett W, Buring JE, Manson JE, Group VR (2020) Effect of vitamin D3 supplements on development of advanced cancer: a secondary analysis of the vital randomized clinical trial. JAMA Netw Open 3(11):e2025850–e2025850

    Article  Google Scholar 

  • Chen W, Clements M, Rahman M, Zhang S, Qiao Y, Armstrong B (2010) Relationship between cancer mortality/incidence and ambient ultraviolet B irradiance in China. Cancer Causes Control CCC 21:1701–9

    Article  Google Scholar 

  • Chowdhury R, Kunutsor S, Vitezova A, Oliver-Williams C, Chowdhury S, Kiefte-de Jong JC, Khan H, Baena CP, Prabhakaran D, Hoshen MB, Feldman BS, Pan A, Johnson L, Crowe F, Hu FB, Franco OH (2014) Vitamin d and risk of cause specific death: systematic review and meta-analysis of observational cohort and randomised intervention studies. BMJ 348:g1903

  • Coren S (1996) Daylight savings time and traffic accidents. New England. J Med 334(14):924–925. PMID: 8596592

    Google Scholar 

  • de Haan K, Groeneveld A, Geus H, Egal M, Struijs A (2014) Vitamin D deficiency as a risk factor for infection, sepsis and mortality in the critically ill: systematic review and meta-analysis. Critical Care (London, England) 18:660

    Article  Google Scholar 

  • Doleac JL, Sanders NJ (2015) Under the cover of darkness: how ambient light influences criminal activity. Rev Econ Stat 97(5):1093–1103

    Article  Google Scholar 

  • Dube A, Lester TW, Reich M (2010) Minimum wage effects across state borders: estimates using contiguous counties. Rev Econ Stat 92 (4):945–964

    Article  Google Scholar 

  • Dudenkov DV, Mara KC, Petterson TM, Maxson JA, Thacher TD (2018) Serum 25-hydroxyvitamin D values and risk of all-cause and cause-specific mortality: a population-based cohort study. Mayo Clin Proc 93(6):721–730

    Article  Google Scholar 

  • Duranton F, Rodriguez-Ortiz M, Duny Y, Rodriguez M, Daures J-P, Argiles A (2013) Vitamin d treatment and mortality in chronic kidney disease: a systematic review and meta-analysis. Amer J Nephrol 37:239–248

    Article  Google Scholar 

  • Edvardsen K, Brustad M, Engelsen O, Aksnes L (2007) The solar UV radiation level needed for cutaneous production of vitamin D3 in the face. A study conducted among subjects living at a high latitude (68cx n). Photochem Photobiol Sci 6:57–62

    Article  Google Scholar 

  • Falagas ME, Karageorgopoulos DE, Moraitis LI, Vouloumanou EK, Roussos N, Peppas G, Rafailidis PI (2009) Seasonality of mortality: the September phenomenon in Mediterranean countries. CMAJ 181(8):484–486

    Article  Google Scholar 

  • Garland CF, Garland FC, Gorham ED, Lipkin M, Newmark H, Mohr SB, Holick MF (2006) The role of vitamin D in cancer prevention. American J Public Health 96(2):252–261. PMID: 16380576

    Article  Google Scholar 

  • Garland CF, Kim JJ, Mohr SB, Gorham ED, Grant WB, Giovannucci EL, Baggerly L, Hofflich H, Ramsdell JW, Zeng K, Heaney RP (2014) Meta-analysis of all-cause mortality according to serum 25-hydroxyvitamin D. American J Public Health 104(8):e43–e50. PMID: 24922127

    Article  Google Scholar 

  • Ginde AA, Liu MC, Camargo J, Carlos A (2009) Demographic differences and trends of vitamin D insufficiency in the US population, 1988-2004. Arch Inter Med 169(6):626–632

    Article  Google Scholar 

  • Hagenau T, Vest R, Gissel T, Poulsen C, Erlandsen M, Mosekilde L, Vestergaard P (2008) Global vitamin D levels in relation to age, gender, skin pigmentation and latitude: an ecologic meta-regression analysis. Osteoporosis international : a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA 20:133–40

    Article  Google Scholar 

  • Hamermesh DS, Myers CK, Pocock ML (2008) Cues for timing and coordination: latitude, letterman, and longitude. J Labor Econ 26 (2):223–246

    Article  Google Scholar 

  • Han J, Guo X, Yu X, Liu S, Cui X, Zhang B, Liang H (2019) 25-Hydroxyvitamin D and total cancer incidence and mortality: a meta-analysis of prospective cohort studies. Nutrients 11(10):2295

    Article  Google Scholar 

  • Havranek T, Herman D, Irsova Z et al (2018) Does daylight saving save electricity? A meta-analysis. Energ J 39(2):35–61

    Article  Google Scholar 

  • Haykal T, Samji V, Zayed Y, Gakhal I, Dhillon H, Kheiri B, Kerbage J, Veerapaneni V, Obeid M, Danish R, Bachuwa G (2019) The role of vitamin D supplementation for primary prevention of cancer: meta-analysis of randomized controlled trials. J Commun Hosp Inter Med Perspect 9:480–488

    Article  Google Scholar 

  • Heaney RP, Davies KM, Chen TC, Holick MF, Barger-Lux MJ (2003) Human serum 25-hydroxycholecalciferol response to extended oral dosing with cholecalciferol 77(1): 204–210

  • Holick MF (2002) Vitamin D: the underappreciated D-lightful hormone that is important for skeletal and cellular health. Current Opin Endocrinol Diab Obes 9(1):87–98

    Article  Google Scholar 

  • Institute of Medicine (2011) Dietary reference intakes for calcium and vitamin D. The National Academies Press, Washington

    Google Scholar 

  • Janszky I, Ljung R (2008) Shifts to and from daylight saving time and incidence of myocardial infarction. New England. J Med 359(18):1966–1968. PMID: 18971502

    Google Scholar 

  • Jin L, Ziebarth N (2020) Sleep, health, and human capital: evidence from daylight saving time. J Econ Behav Organ

  • Joergensen C, Gall M-A, Schmedes A, Tarnow L, Parving H-H, Rossing P (2010) Vitamin D levels and mortality in type 2 diabetes. Diabetes Care 33(10):2238–2243

    Article  Google Scholar 

  • Joergensen C, Hovind P, Schmedes A, Parving H-H, Rossing P (2011) Vitamin D levels, microvascular complications, and mortality in type 1 diabetes. Diabetes Care 34(5):1081–1085

    Article  Google Scholar 

  • Kellogg R, Wolff H (2007) Does extending daylight saving time save energy? Evidence from an australian experiment. IZA Discussion Papers 2704. Institute of Labor Economics (IZA)

  • Khan QJ, Fabian CJ (2010) How i treat vitamin d deficiency. J Oncol Pract 6(2):97–101. PMID: 20592785

    Article  Google Scholar 

  • Kimlin MG, Olds WJ, Moore MR (2007) Location and vitamin D synthesis: is the hypothesis validated by geophysical data? J Photochemis Photobiol B: Biol 86(3):234–239

    Article  Google Scholar 

  • Kotchen MJ, Grant LE (2011) Does daylight saving time save energy? Evidence from a natural experiment in indiana. Rev Econ Stat 93(4):1172–1185

    Article  Google Scholar 

  • Kroll M, Bi C, Garber C, Kaufman H, Liu D, Caston-Balderrama A, Zhang K, Clarke N, Xie M, Reitz R, Suffin S, Holick M (2015) Temporal relationship between vitamin D status and parathyroid hormone in the United States. PloS one 10:e0118108

    Article  Google Scholar 

  • Licher S, de Bruijn RF, Wolters FJ, Zillikens MC, Ikram MA, Ikram MK (2017) Vitamin D and the risk of dementia: the Rotterdam study. J Alzheimers Dis 60(3):989–997

    Article  Google Scholar 

  • Liu C, Politch JA, Cullerton E, Go K, Pang S, Kuohung W (2017) Impact of daylight savings time on spontaneous pregnancy loss in in vitro fertilization patients. Chronobiol Int 34(5):571–577. PMID: 28156172

    Article  Google Scholar 

  • Manning WG, Mullahy J (2001) Estimating log models: to transform or not to transform? J Health Econ 20(4):461–494

    Article  Google Scholar 

  • Martineau AR, Jolliffe DA, Hooper RL, Greenberg L, Aloia JF, Bergman P, Dubnov-Raz G, Esposito S, Ganmaa D, Ginde AA, Goodall EC, Grant CC, Griffiths CJ, Janssens W, Laaksi I, Manaseki-Holland S, Mauger D, Murdoch DR, Neale R, Rees JR, Simpson S, Stelmach I, Kumar GT, Urashima M, Camargo CA (2017) Vitamin D supplementation to prevent acute respiratory tract infections: systematic review and meta-analysis of individual participant data. BMJ 356

  • Mathyssen C, Gayan-Ramirez G, Bouillon R, Janssens W (2017) Vitamin D supplementation in respiratory diseases - evidence from RCT. Polish Arch Int Med 127

  • Pilz S, Iodice S, Zittermann A, Grant W, Gandini S (2011) Vitamin D status and mortality risk in CKD: a meta-analysis of prospective studies. Amer J Kidney Dis Off J Nat Kidney Found 58:374–82

    Article  Google Scholar 

  • Prerau D (2005) Seize the daylight: the curious and contentious story of daylight saving time. Thunder’s Mouth Press

  • Samji V, Haykal T, Zayed Y, Gakhal I, Veerapaneni V, Obeid M, Kheiri B, Badami S, Bachuwa G, Danish R (2019) Role of vitamin D supplementation for primary prevention of cancer: meta-analysis of randomized controlled trials. J Clin Oncol 37(15):1534–1534

    Article  Google Scholar 

  • Santos Silva J, Tenreyro S (2011) Further simulation evidence on the performance of the Poisson pseudo-maximum likelihood estimator. Econ Lett 112 (2):220–222

    Article  Google Scholar 

  • Schleck M-L, Souberbielle J-C, Jandrain B, Da Silva S, De Niet S, Vanderbist F, Scheen A, Cavalier E (2015) A randomized, double-blind, parallel study to evaluate the dose-response of three different vitamin D treatment schemes on the 25-hydroxyvitamin D serum concentration in patients with vitamin D deficiency. Nutrients 7:5413–5422

    Article  Google Scholar 

  • Silva JS, Tenreyro S (2006) The log of gravity. Rev Econ Stat 88(4):641–658

    Article  Google Scholar 

  • Smith AC (2016) Spring forward at your own risk: daylight saving time and fatal vehicle crashes. Amer Econ J Appl Econ 8(2):65–91

    Article  Google Scholar 

  • Tealde E (2021) The unequal impact of natural light on crime. J Popul Econ

  • Toro W, Tigre R, Sampaio B (2015) Daylight saving time and incidence of myocardial infarction: evidence from a regression discontinuity design. Econ Lett 136:1–4

    Article  Google Scholar 

  • Wacker M, Holick M (2013) Vitamin D effects on skeletal and extraskeletal health and the need for supplementation. Nutrients 5:111–48

    Article  Google Scholar 

  • Wang Y, Yang Z, Gao L, Cao Z, Wang Q (2020) Effects of a single dose of vitamin D in septic children: a randomized, double-blinded, controlled trial. J Int Med Res 48(6):0300060520926890. PMID: 32485124

    Article  Google Scholar 

  • Webb A, Kline L, Holick MF (1988) Influence of season and latitude on the cutaneous synthesis of vitamin D3: exposure to winter sunlight in Boston and Edmonton will not promote vitamin D3 synthesis in human skin. J Clin Endocrinol Metabol 67(2):373–378

    Article  Google Scholar 

  • Wickham R (2012) Cholecalciferol and cancer: is it a big D3-eal? J Adv Pract Oncol 3:249–57

    Google Scholar 

  • Wolff H, Makino M (2012) Does daylight saving time burn fat? Time allocation with continuous activities

  • Xu Y (2017) Generalized synthetic control method: causal inference with interactive fixed effects models. Polit Anal 25(1):57–76

    Article  Google Scholar 

  • Yetley EA (2008) Assessing the vitamin D status of the US population. Amer J Clin Nutrit 88(2):558S–564S

    Article  Google Scholar 

  • Zadshir A, Tareen N, Pan D, Norris K, Martins D (2005) The prevalence of hypovitaminosis D among US adults: data from the NHANES III. Ethn Dis 15(4):S5

    Google Scholar 

  • Zhang Y, Fang F, Tang J, Jia L, Feng Y, Xu P, Faramand A (2019) Association between vitamin D supplementation and mortality: systematic review and meta-analysis. BMJ 366

Download references

Acknowledgements

I thank the editor, Dr. Klaus F. Zimmermann, and the four anonymous referees, whose comments and guidance substantially improved this study. Finally, special thanks to Michele Starkey of the Indiana State Department of Health, without whom this study would have never been possible.

Author information

Authors and Affiliations

  1. SUNY at Fredonia, Fredonia, NY, 14063, USA

    Adam Cook

Authors
  1. Adam Cook
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Adam Cook.

Ethics declarations

Conflict of Interest

The author declares no competing interests.

Additional information

Responsible editor: Klaus F. Zimmermann

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix 1

Appendix 1

1.1 Collapsing the time series information

To control for the possibility that the significance of the results is driven by autocorrelation bias, the month-year fixed effects in the regression equation are collapsed into pre- and post-treatment time periods as described in section IV.C of Bertrand et al. (2004). When the time series information is collapsed into pre- and post-treatment periods, as expressed in Eq. 2, autocorrelation between time periods is no longer a concern.

$$ \begin{array}{@{}rcl@{}} log(y_{it}) & = & \alpha+\beta POSTEXP+\gamma_{i}COUNTY_{i} \\ & & +\eta_{1}ADOPTER_{i}+\eta_{2}DST_{t}+\eta_{3}(POSTEXP_{t}\times DST_{t}) \\ & & +\eta_{4}(ADOPTER_{i}\times POSTEXP_{t})+\eta_{5}(ADOPTER_{i}\times DST_{t}) \\ & & +\delta(ADOPTER_{i}\times POSTEXP_{t}\times DST_{t})+X_{it}^{T}\xi_{it}+\mu_{it} \end{array} $$
(2)

The all-cause mortality results using collapsed time series information are reported in Table 7.

Table 7 Collapsed time series information all-cause mortality results
Full size table

The collapsed time series DST treatment effects are significant and comparable to the DST treatment effects estimated in the main results in Table 2.Footnote 29 The results using collapsed time series information indicate that autocorrelation bias is likely not responsible for the significance of the DST treatment effects when year-month fixed effects are included.

1.2 Specific-cause results, quasi-Poisson regression

One concern when implementing a Poisson regression model is the validity of the Poisson variance assumption: Unlike linear regression which requires the variance to be constant, the Poisson variance assumption requires that the variance of the dependent variable equals the mean of the dependent variable.Footnote 30 To guard against this possibility, I replicate the results of the Poisson regression model using Poisson quasi-maximum likelihood estimation, which allows for deviations from the restrictive Poisson variance assumption. The dispersion parameter (τ) reflects the magnitude of over- or under-dispersion present in the model. Dispersion parameters greater than 1 imply overdispersion, when the variance exceeds the mean, and dispersion parameters less than 1 imply a mean larger than the variance. I report the quasi-Poisson results in Table 8 along with the dispersion parameter for each quasi-Poisson specific-cause treatment effect estimation to support using Poisson regression models to estimate the cause-specific treatment effects.

Table 8 Quasi-Poisson estimates of DST specific-cause mortality effects
Full size table

The quasi-Poisson estimates and standard errors correspond closely with the Poisson regression estimates, and for all estimations, the dispersion parameters are all approximately 1, supporting the validity of the Poisson variance assumption.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Cook, A. Saving lives: the 2006 expansion of daylight saving in Indiana. J Popul Econ 35, 861–891 (2022). https://doi.org/10.1007/s00148-021-00833-6

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00148-021-00833-6

Keywords

JEL Classification

Search

Navigation